High Pressure 450 bar Natural

Gas Injection Compressor

HITOSHI SHINOHARA* 1
AKINORI TASAKI* 1
JUNICHI HORIBA* 1
JYOU MASUTANI* 2

As the demand for oil and gas increases, the need for enhanced oil recovery, also known as EOR, and the demand
of natural gas injection compressors has been on the rise. Mitsubishi Heavy Industries, Ltd. (MHI) has developed an
injection compressor placing particular emphasis on reliability for handling high-pressure and high-density gas in
addition to high efficiency as essential features for MHIs compressors. MHI carried out shop full load tests using
natural gas at a discharge pressure of 450 bar, and demonstrated the high reliability of the compressor under the
same operating conditions as those on site.

The impeller was divided into high- and low-pressure

sections, and arranged as back to back. Thus, losses due
to internal circulation flow could be reduced to increase
efficiency, and variations in shaft thrust force due to variations in operating conditions could also be reduced. When
the impellers are arranged in series, the thrust force may
vary significantly. Continuous operation may be obstructed when the compressor is operated at off-design
points or the labyrinth seal is worn by long-term operation. However, the thrust forces are set against each other
through a back-to-back arrangement so as to reduce variations in the total thrust force acting on the shaft.

1. Introduction
Natural gas injection compressors are used for reinjecting gas deep in the ground near oil wells and
natural gas wells. Since they handle ultra high-pressure and high-density gas, they have unique difficulties
in terms of design and operation. In actual operations
on site, a large number of problems with compressors,
including the unstable shaft vibration, have also been
reported.
MHI has delivered a variety of high-pressure compressors mainly for the petrochemical industry, and has
developed a new natural gas injection compressor using
technologies developed so far. In the design of the compressor, emphasis was placed mainly on ensuring high
reliability demanded under a wide range of use conditions.

2. Specifications of compressor
In some cases, compressors having a discharge pressure of up to 700 bar are used for natural gas injection.
Since the frequency of use of compressors with discharge
pressures of 450 bar or less is high, MHI used this pressure as a design parameter. The specifications of the
newly developed compressor are shown in Table 1
1, and a
Fig.
1
cross sectional view is shown in
1.

Table 1 Specifications of injection compressor

Compressor model
Gas handled
Suction pressure
Discharge pressure
Design speed
Required power

3V-6B

High rigidity rotor

High bore / tip ratio impeller
350 bar tandem dry gas seal
Abradable seal with swirl canceller
Reduction in flow at point where rotating swirl occurs
High precision gas passage impeller

Natural gas
205 bar
450 bar
11 000 rpm
8 900 kW

Fig. 1 Cross sectional view of injection compressor

Mitsubishi Heavy Industries, Ltd.

Technical Review Vol. 41 No. 3 (Jun. 2004)

*1 Hiroshima Machinery Works

*2 Takasago Research & Development Center, Technical Headquarters

In addition, the shaft diameter and bearing diameter

relative to the diameter of the impellers were increased,
and the bearing span was minimized. Use of this highly
rigid rotor has made it possible to realize an increase in
the reliability on high-density gas with respect to instability.

The design values were found to match the predicted

values quite well. As a result, the techniques used for
predicting performance were confirmed to be highly accurate, and the test and evaluation methods based on
the law of aerodynamic similarity were also verified to
be appropriate for application to validating high-pressure compressor designs, as well.
To confirm the performance of the compressor under
actual operating conditions, full load performance tests
were performed using natural gas at a discharge pressure of 450 bar. The test conditions, based on ASME
PTC-10 type I, are equivalent to the actual operating
conditions on site. The relation between the flow coefficient obtained from the tests and the polytropic
efficiency/work coefficient is shown in Fig. 3
3.

3. Aerodynamic performance

: Low pressure test

: Actual load test

0.8
0.7
0.6
0.5
0.4
0.3
0.2

Pressure fluctuation
coefficient

Polytropic work coefficient

Polytropic efficiency

3.1 Design and manufacture

The aerodynamic design and manufacture of the impeller can be briefly summarized as follows.
(1) A high efficiency three-dimensional impeller is applied to all compressors.
(2) All standard impellers are designed using CFD
analysis.
(3) Performance tests are performed on all standard impellers, and a database for the results thus obtained
is built and maintained.
(4) All impellers are similarly designed based on data on
the standard impellers. Verified data is also used for
the physical properties of the gas, which is indispensable to proper aerodynamic design.
(5) All impellers are machined using highly accurate, 5axis control machine.
3.2 Results of performance tests
Shop performance tests were conducted in accordance
with the ASME PTC-10 Performance Test Code to verify
the validity of the design of the injection compressor in
terms of performance. The relation between the flow coefficient obtained from the tests and the polytropic
efficiency/work coefficient is shown in Fig. 2
2.

34%

0.03
0.02
0.01

0.005

0.010

0.015

0.020

0.025

Flow coefficient
: 100 % rotational speed measuring point
: 105 % rotational speed measuring point
: Sect. 1 100 % predicted curve
: Sect. 2 100 % predicted curve
: Sect. 1 105 % predicted curve
: Sect. 2 105 % predicted curve

(a) Section 1 test results

0.8

Relative polytropic work coefficient

Surge measurement point

1.0
0.8
0.6
0.4

0.6
0.5
0.4
0.3
0.2

Sect. 2 Sect.1
0.2

0.010

22%

0.005

0.005
3V-6B

0.0
0.00

0.7

Pressure fluctuation
coefficient

Polytropic work coefficient

Polytropic efficiency

1.2

0.01

0.010

0.015

0.020

Flow coefficient
0.02

0.03

Flow coefficient
Fig. 2 Results of low pressure performance test
Calculated values indicated by solid and dotted lines match well
the measured values, which are indicated by
marks.

(b) Section 2 test results

Fig. 3 Comparison between low pressure test results and full load
performance test results
Low pressure test results indicated by marks closely match the full load
performance test results indicated by marks.

Mitsubishi Heavy Industries, Ltd.

Technical Review Vol. 41 No. 3 (Jun. 2004)

The results of the full load tests also match the

results of the low-pressure tests, thereby confirming
the predictive accuracy of performance under high
pressure.
3.3 Rotating stall
It is well known that when gas flow is reduced from
normal operating conditions, gas flow around the impeller and diffuser exhibits rotating stall before surging.
Though this occurs in all compressors, in many cases, it
is not detected from outside the compressor and does not
cause any problems with operation.
However, since high-density gas is used in an injection compressor, the compressor generates unstable shaft
vibrations as a result of the gas force that is produced
by rotating stall. Such instability could potentially lead
to trouble during continuous operation.
A diffuser with a smaller width than that normally
used in a low-pressure compressor was adopted in the
injection compressor. As a result, the point at which rotating stall occurs can be extended to a low flow.
Fig. 3 shows the results obtained by measuring pressure fluctuations caused by rotating stall in full load
performance tests with a sensor fixed on the diffuser.
Rotating stall occurs at low flow point of 34% in the
first section from the design point and at a low flow
point of 22% in the second section from the design point.
This indicates that a safe operating range has been sufficiently secured.

4. Rotor dynamics
4.1 Critical speed
As can be seen in Fig. 4
4, the calculated critical speed
of the compressor sufficiently differ from the operating
speed, and the amplification factor is also small. The results obtained from the shop tests match the calculated
values quite well, thereby confirming the validity of the
model used for the calculations.
4.2 Stability of rotor
When the excitation force acting on the rotor is increased compared with the damping force, an unstable
self-exciting vibration occurs in the rotor. In a compressor handling high-pressure gas, the exciting force
is produced mainly by gas swirl at the labyrinth seal
section. The smaller the labyrinth clearance, the
larger the exciting force becomes. On the other hand,
a damping force is generated mainly at the bearing
section.
An abradable seal is used at the labyrinth section of
an injection compressor to increase aerodynamic performance. However, in order to minimize the excitation
force, a swirl canceller exclusive to MHI was installed to
prevent gas swirl. In addition, an MHI exclusive overhung damper was also applied in order to increase the
damping force. This is formed by adding a shaft vibration damping function formed by an oil film to the
rotating shaft section.

Vibration mode

Amplitude (cm)

Amplitude (cm)

Phase (deg)

Mode circle

Position on model

Rotational speed (rpm)

Results of calculation
Rotational
Amplification Rotational
Amplification
speed (rpm) Mode
speed (rpm) Mode
factor
factor
4 650

H-1st

1.5

19 000

H-2 st

4.2

5 000

V-1st

3.3

19 000

V-2 st

5.1

Measured values in shop test

Rotational
Amplification
Mode
speed (rpm)
factor
5 100

V-1 st

2.6

Fig. 4 Analysis of critical speed

The calculated critical speed match well with the actually measured values, and the amplification factor is also small.

Mitsubishi Heavy Industries, Ltd.

Technical Review Vol. 41 No. 3 (Jun. 2004)

N 1N 2 N

4N

8N

N 1N 2 N

400

800

1 200

1 600

2 000

20
10

8N

400

800

1 200

1 600

2 000

400

800

1 200

1 600

2 000

400

800

1 200

1 600

2 000

400

800

1 200

1 600

2 000

20
10
400

800

1 200

1 600

2 000

Amplitude

Amplitude

4N

20
10

20
10

20
10
400

800

1 200

1 600

2 000

0
20
10
0

20
10

20
10
400

800

1 200

1 600

2 000

Frequency (Hz)

Frequency (Hz)

(a) No-load test (rotational speed: 12 999 rpm)

(b) Full load test (rotational speed: 11 362 rpm)

Fig. 5 Analysis of shaft vibration frequency

In both the low-pressure test and full load test, the nonsynchronous components of the operational speed are sufficiently small.

Accelerometer

14

Load weight

12

Exciter

Damping ratio (%)

Hydraulic system

Accelerometer

Compressor

S/T
Drive
machine

10
8
6
4

: Measured values
: Calculated values

Excitation
force value
Analyzer

Excitation
signal

Shaft
vibration
value

2
0

5 000

10 000

Compressor load (kW)

(a) Excitation test block diagram

(b) Rotor damping ratio

Fig. 6 Results of compressor excitation test

The excitation test was performed as part of the full load test. As a result, the damping
ratio was found to be 8.7%. This indicates that the compressor is sufficiently stable.

tion due to centrifugal force is optimized, the disk is

joined to the cover by welding or diffusion welding, and
the impeller is formed so as to be shrinkage-fitted to the
shaft.
Since the natural frequency of the impeller is lowered in a high-density gas environment, the design of it
was formed so as to avoid it to secure reliability.
5.2 Bearings
Shaft thrust force is reduced by the back-to-back arrangement of the impeller, and a direct lubrication
system is applied to the thrust bearings and journal bearings to increase load capacity. The thrust collar is
hydraulically fitted to the shaft to increase reliability
and facilitate maintenance.

The results of analysis of the shaft vibration frequency

measured during in shop tests are shown in Fig.5
Fig.5. It
could confirmed that, under full load operating conditions, any vibration of nonsynchronous components that
may cause unstable vibration does not occur also as in
the no-load operation.
In addition, the compressor excitation test as shown
in Fig. 6 was performed, which confirmed that the rotor
has sufficient damping force and that stability has been
secured.

5. Mechanical design
5.1 Impeller
The impeller is formed in such a way that deforma-

Mitsubishi Heavy Industries, Ltd.

Technical Review Vol. 41 No. 3 (Jun. 2004)

100

Allowable metal temperature 120 C

: Thrust bearing metal temperature
: Journal bearing metal temperature
: Shaft vibration

800

Allowable shaft vibration 25.4

30

40

20

20

10

Leakage ( L / min)

80
60

Allowable leakage at 350 bar

700
Shaft vibration

Bearing metal temperature ( C)

120

: Drive machine side

: Anti-drive machine side
Test rotational speed 13 500 rpm

600
500
400
300
200

Leak amount in
actual load test

100

100

200

50

100

150

200

250

300

350

400

Seal pressure (bar)

Suction pressure (bar)

Fig. 7 Bearing metal temperature and shaft vibration amplitude
Both the metal temperature and shaft vibration amplitude are stable.

Fig. 8 Results of measurement of dry gas seal leakage

Fig. 7 shows bearing metal temperatures measured

during the full load tests together with the shaft vibration amplitude. All values were sufficiently small against
the allowable values. In addition, it was also confirmed
that variation of these values due to variations in operating pressure are small, as well as stable under
high-pressure operation.
5.3 Seals
A tandem dry gas seal with a design pressure of 350
bar was applied to the shaft seal. Compressor discharge
gas is supplied as seal gas through a filter. Fig. 8 shows
that the amount of leakage from the seal during the full
load tests matches well the results of the bench test. This
indicates the integrity of the seal in the full load tests.

6. Conclusion
MHI has developed a new natural gas injection compressor, putting emphasis on reliability under the specific
operating conditions of the unit.
Using natural gas as a test gas, full load shop tests
were conducted at a discharge pressure of 450 bar
which reproduces the same operating conditions as
those on site. Based on the results of the tests, it was
verified that the design and manufacture of the compressor were appropriate with respect to the
mechanical properties including shaft vibration and
aerodynamic performance, resulting in the compressor having high reliability.

Hitoshi Shinohara

Akinori Tasaki

Junichi Horiba

Jyou Masutani

Mitsubishi Heavy Industries, Ltd.

Technical Review Vol. 41 No. 3 (Jun. 2004)